Journal of Behaviour Therapy

and Experimental Psychiatry 31 (2000) 259–279

Imaginal exposure for anger reduction in adult

outpatients: a pilot study
Gustavo R. Grodnitzkya, Raymond Chip Tafrateb,*
a
Adirondack Samaritan Counseling Center, Hudson Falls, New York, USA
b
Central Connecticut State University, Department of Criminology and Criminal Justice,
1615, Stanley Street, New Britain, CT 06050, USA
Received 24 March 2000; received in revised form 29 March 2001; accepted 18 April 2001

Abstract

Although exposure procedures have been widely accepted in the treatment of anxiety
disorders, they have rarely been applied to the treatment of anger. The present paper describes
an initial attempt to apply an imaginal exposure strategy to adult outpatients (n ¼ 6) referred
for anger management. This investigation reflects an empirical clinical practice approach
rather than a controlled outcome study. Thus, this paper provides a clinical description of the
imaginal exposure program, pre-to-posttest effectiveness data, an exploration of habituation
patterns for each participant, and 15-month follow-up data from several patients. In
considering the impact of the intervention, statistically significant change was found on most
anger variables, the majority of patients met a criteria for clinically significant improvement on
important indices of anger, and treatment effect sizes were large and compared favorably to
previously studied interventions. Process data revealed a consistent habituation effect, across
patients and anger stimuli, in response to repeated exposure practice. Participants’ satisfaction
was also positive. Finally, statistically significant and clinically meaningful change was evident
at 15-months following the intervention. Data from the current pilot project are encouraging
and hopefully will stimulate more methodologically rigorous clinical trials. r 2001 Elsevier
Science Ltd. All rights reserved.

Although anger is an emotional problem frequently encountered in clinical

practice (Lachmund a DiGiuseppe, 1997), its nature and treatment have received
relatively little attention in the scientific literature (Kassinove a Sukhodolsky,
1995). As a clinical problem, anger has been associated with a number of serious
negative consequences such as aggressive behavior (Cornell, Peterson, a Richards,

*Corresponding author. Tel.: +1-860-832-3147.

E-mail address: tafrater@ccsu.edu (R.C. Tafrate).

0005-7916/01/$ - see front matter r 2001 Elsevier Science Ltd. All rights reserved.
PII: S 0 0 0 5 - 7 9 1 6 ( 0 1 ) 0 0 0 1 0 - 6
260 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279

1999; Deffenbacher, Oetting, Lynch, a Morris, 1996), family violence (Jacobson

et al., 1994; Whiteman, Fanshel, a Grundy, 1987), substance abuse (DeMoja a
Spielberger, 1997; Walfish, Massey, a Krone, 1990), and physical health problems
(Helmers, Posluszny, a Krantz, 1994). It is therefore important that effective
interventions be developed and studied. Unfortunately, at present, only a few
treatment modalities have obtained any degree of empirical support. These
modalities include relaxation, cognitive, and skills training interventions, as well as
combinations of these three approaches. For quantitative reviews of the adult
treatment outcome literature see Edmondson and Conger (1996) and Tafrate
(1995). In addition to the small number of treatment outcome studies and the
limited breadth of strategies examined, the majority of investigations have relied
exclusively on undergraduate student volunteers (Tafrate, DiGiuseppe, a Parsi,
1997). This of course raises concerns about whether findings obtained in university
settings, with students samples, can be generalized to the types of patients seeking
anger treatment in everyday clinical practice or other types of service delivery
settings.
The present study sought to evaluate the effectiveness of a novel treatment for
anger in a small sample of patients referred to a community mental health center.
Although exposure has been widely accepted as an active therapeutic element in
treatment programs for anxiety disorders (Barlow a Craske, 1994; Craske, Barlow,
a O’Leary, 1992; Foa a Rothbaum, 1998; Foa a Wilson, 1991; Markway,
Carmin, Pollard, a Flynn, 1992), it has rarely been conceptualized and applied as
an active intervention for anger reduction. However, there have been several
investigations that have incorporated exposure procedures into treatment programs
for angry subjects. In an investigation designed to evaluate the effectiveness of
meditation and negative thought reduction for angry college students, Dua and
Swinden (1992) utilized an unusual placebo condition in which subjects imagined
high anger producing situations. It was found that the placebo (imaginal exposure)
procedures were as effective in reducing anger as the other active interventions.
Perhaps since this approach was considered to be a placebo, little information was
provided on how the exposure intervention was implemented, and there was no
discussion regarding reactions and habituation patterns of subjects.
In another treatment study, Tafrate and Kassinove (1998) exposed self-referred
angry men to anger-provoking verbal barbs (in-vivo exposure) while they rehearsed
rational (based on Ellis’ Rational Emotive Behavior Therapy [REBT]; Ellis, 1962),
irrational, or irrelevant self-statements. Although subjects who rehearsed self-
statements based on the REBT model showed the greatest improvement, significant
pre-to-posttest anger reduction was also observed for subjects in the other two
groups. While the barb exposure was hypothesized to be the common effective
treatment ingredient across all conditions, this study had not been designed to
directly evaluate the effects of the exposure component. Finally, imaginal exposure
procedures have been incorporated into relaxation interventions intended for angry
subjects. For example in anger management training (AMT; Hazaleus a Deffen-
bacher, 1986; Suinn, 1990) relaxation skills are practiced in response to anger scenes
developed by the client. Several meta-analytic reviews compared AMT to progressive
 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279 261

muscle relaxation alone and noted increased effectiveness in protocols that included
imaginal anger scenes (Tafrate et al., 1997; Tafrate, 1995).
Several studies have also examined the effectiveness of systematic desensitization
with angry clients. Standard systematic desensitization (relaxation and exposure to
imaginal scenes) has been found to be effective for male college students who
reported high anger while driving (Rimm, DeGroot, Boord, Heiman, a Dillow,
1971), white male college students who reported racial anger (O’Donnell a Worell
1973), and student nurses (Evans, Hearn, a Saklofske, 1973). In the O’Donnell and
Worell study, it was also found that desensitization alone (in the absence of
relaxation) was not effective. It is possible that the short number of exposure trials
(five) was not sufficient to achieve habituation of anger responses. While results
from these studies are encouraging, the use of student samples and the lack of
standardized and rigorous anger measures raise questions about the effectiveness of
systematic desensitization with clinical patients. Regrettably, no additional
published research on this approach for anger has appeared since the early 1970s.
There has also been little written about the theoretical underpinnings of an
exposure model of anger treatment. It has been proposed that the underlying
mechanisms would be similar to those believed efficacious in the treatment of anxiety
disorders (Brondolo, DiGiuseppe, a Tafrate, 1997). Clients with anger problems
often report automatic-like responses to anger stimuli. It is conceivable that some
types of anger problems arise through classical conditioning and are subsequently
maintained through operant conditioning. Behaviors such as arguing, blaming
others, and aggressive actions may result in temporary positive feelings and
compliance by others. Such behaviors also constitute avoidance of negative emotions
and do not allow for anger to extinguish. Repeated and prolonged exposure to an
anger-evoking trigger, while preventing these usual response patterns, will interrupt
and weaken the chain-of-events (perceptions, cognitions, physical arousal, a
behaviors) linking a trigger to a response. In addition, exposure to the emotion
itself, independent of the provocation, may also be useful in reducing feelings
frequently associated with anger such as fear, resentment, shame, and hurt. As
individuals habituate to their triggers they may become better able to tolerate the
experience of anger and develop more flexible cognitive and behavioral responses to
provocation.
The present study was undertaken to yield pilot data on imaginal exposure applied
as an active and structured intervention, administered to a clinical sample, in a
common service delivery setting. A secondary goal was to determine if the process of
habituation in angry clients is similar to what has been observed in clients with
anxiety disorders. The present study utilized a pre-to-posttest design with 6 subjects.
The choice of design was guided less by stringent experimental considerations and
more by real world clinical concerns (e.g. withholding of treatment for the purposes
of establishing a comparison group or collecting baseline data was not considered
feasible). Due to the small sample size and clear limitations inherent in the design,
several approaches to judging the effectiveness of the intervention were selected
including tests of statistical significance, an index of clinically significant and reliable
change, effect size estimates, process data, and 15-month follow-up. For this reason
262 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279

the current project represents an empirical clinical practice approach rather than the
more traditional clinical trial.

1. Method

1.1. Participants

Participants were drawn from a pool of 25 referrals made to a community mental

health center for anger management. Eligibility for the project was restricted to
individuals over 18, who scored in the upper quartile (scores>21) on the Trait Anger
Scale (Spielberger, 1988). Eleven of the referrals had anger control problems that
occurred exclusively during domestic disputes and were directed to an already
existing domestic violence program. Six of the referrals did not meet the anger
criteria, and two preferred individual treatment. Thus, 6 adults (five males and one
female) participated in the present program. Four of these patients were referred to
the center by the county Court system, one by another mental health professional,
and one client was self-referred.
The level of anger reported by the patients, as measured by pretest means on the
Trait Anger Scale (Spielberger, 1988), indicated that this sample was angrier than
95% of normal adults (M ¼ 27:00; SD ¼ 4:56). Levels of anger were also higher
than those reported for male prison inmates (Kroner a Reddon, 1995) and similar
to treatment samples of angry adults (Tafrate a Kassinove, 1998) and college
students (Deffenbacher a Stark, 1992; Deffenbacher, Thwaites, Wallace, a
Oetting, 1994). The ages of the patients ranged from 24 to 45 (M ¼ 34:33;
SD ¼ 8:94). Five were divorced, one was separated, and five had children. All were
employed full-time and their self-reported yearly incomes ranged from $18,000 to
$65,000 (M ¼ 32:67; SD ¼ 18:95). All participants were Caucasian and their years of
education ranged from 12 to 16 (M ¼ 13:67; SD ¼ 1:50). Although the present
program is focused on the emotion of anger (as a phenomenological/internal
experience), four of the patients had been arrested at least once for aggressive
behaviors such as brawling and assault. All reported significant negative
consequences associated with anger related episodes (e.g. damage to family
relationships, loss of romantic relationships, interference with work).

1.2. Measures

1.2.1. Norm-based anger measures

In order to make comparisons to other investigations in the treatment outcome
literature and to calculate indices of reliable and clinical improvement (see results),
portions of the State-Trait Anger Expression Inventory (Spielberger, 1988) were
administered to patients at pretest and posttest. The Trait Anger Scale is designed to
measure an individual’s propensity to experience and express anger. This scale
consists of 10 statements that describe subjective feelings of anger. In response to the
sentence stem, ‘‘How I generally feel’’, participants rated on a 4-point Likert type
 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279 263

scale (1=almost never to 4=almost always) how characteristic each item was for
them. Higher scores indicate greater general anger. This scale has been shown to be
internally consistent (a ¼ 0:82; Spielberger, 1988), to correlate positively with other
measures of anger (Spielberger, 1988), and to discriminate high anger individuals
from others (Deffenbacher, Demm, a Brandon, 1986; Lopez a Thurman, 1986).
Modes of anger expression (e.g. the tendency to generally hold anger in, to express
anger outwardly, or to remain calm and control anger) were assessed by the Anger
Expression Scale. This self-report measure consists of three 8-item subscales (anger-
in, anger-out, and anger-control) on which participants were asked to rate, on a 4-
point Likert type scale (1=almost never to 4=almost always), the degree to which
each statement described how they expressed themselves when angry. Higher scores
reflect greater tendency to engage in that particular mode of expression. The anger
expression scales have internal consistency reliabilities that range from 0.73 to 0.84,
have been shown to correlate positively with other measures of anger (Deffenbacher,
1992; Spielberger, 1988), and are not highly positively correlated with each other
(Spielberger, 1988).

1.2.2. Idiographic anger measures

In order to obtain a measure of each participant’s anger in relation to a specific
real life situation, an idiographic assessment strategy was adapted from Hazaleus
and Deffenbacher (1986) and more recently Deffenbacher, Dahlen, Lynch, Morris,
and Gowensmith (2000). Each participant was asked to identify the ongoing
situation associated with the greatest feeling of anger. In response to that situation,
ratings were obtained on the following anger dimensions at pretest and posttest:
(a) anger intensity (0=little or no anger, 100=maximum level of anger ever
experienced), (b) frequency (number of times anger had been experienced in that
situation in the past month), (c) duration (the number of minutes that anger usually
lasted), (d) the degree to which anger in the situation interfered with the persons
functioning (0=no interference, 100=extreme interference), and (e) the intensity of
the strongest anger related physical sensation experienced in the situation (0=no
symptoms, 100=extremely severe symptoms). As reported by Deffenbacher et al.
(1996), intensity ratings for the most salient ongoing anger situation have a 10-week
test-retest reliability of 0.81 and correlate positively with other anger measures.

1.2.3. Process measures

The Daily Anger Exposure Record was created by the authors as a method of
structuring and documenting patients exposure practice.1 For each exposure trial,
patients indicated the anger scene content, the time that the exposure trial began and
ended, physical sensations that occurred during exposure, and anger intensity ratings
in response to the scene (0=none, 8=extreme). Anger intensity ratings are similar to
the subjective units of distress scale (SUDS) commonly used in imaginal exposure
1
Requests for copies of the Daily Anger Exposure Record should be addressed to Raymond Chip
Tafrate, Ph.D., Department of Criminology a Criminal Justice, Central Connecticut State University,
1615 Stanley Street, New Britain, CT 06050; e-mail: tafrater@ccsu.edu.
264 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279

procedures for clients with anxiety disorders. Patients provided anger ratings at two
points during each trial, after the first minute and at the end of the one half-
hour trial.
Since reports of imaginal exposure are relatively new to the anger treatment
literature, it was unclear how patients would react to the procedures. Potential
concerns included patients not understanding the rationale for exposure practice,
lack of rapport with the group leader, and dissatisfaction with the treatment pro-
tocol. Therefore, the client form of the Working Alliance Inventory (WAI; Horvath
a Greenberg, 1989) was administered at posttest. This 36-item questionnaire
contains three subscales; clients’ perception of agreement on goals, perception of
agreement on tasks, and the formation of a bond. Participants rated on a Likert type
scale (1=never to 7=always) the degree to which each item reflected the alliance
between themselves and the group leader during the course of treatment. Alpha
reliabilities have been reported to range from 0.75 to 0.91 for an adult sample of
angry men, suggesting good internal consistency (Tafrate a Kassinove, 1998). Low
scores on the WAI have been associated with poor outcome and premature
termination while higher scores have been associated with good treatment outcome
(Samstag, Batchelder, Muran, Safran, a Winston, 1998).

1.2.4. Norm-based measures of other emotional disturbance

Measures of anxiety and depression were also included to assess the potential
impact of treatment on non-targeted emotional issues. Decreases in both anxiety and
depression have been noted in several anger studies (Deffenbacher et al., 2000;
Deffenbacher a Stark, 1992). On the 21-item Beck Depression Inventory-Second
Edition (BDI-II; Beck, Steer, a Brown, 1996) participants rated the severity of
depressive symptoms experienced during the previous week. The BDI-II has alpha
coefficients that range from 0.92to 0.93 and correlates positively with the original
BDI, other measures of depression, as well as measures of hopelessness and suicidal
ideation (Beck et al., 1996). On the 21-item Beck Anxiety Inventory (BAI; Beck a
Steer, 1993), participants rated the severity of cognitive and somatic symptoms of
anxiety experienced during the previous week. Alpha coefficients for the BAI range
from 0.92 to 0.94 and the BAI has been shown to correlate positively with other
measures of anxiety (Beck a Steer, 1993).

1.3. Procedure

Individuals referred to the clinic for anger management met individually with the
first author for an initial screening session. The Trait Anger Scale was administered
and those who met the criteria for participation completed an intake form, an
informed consent form, and all additional anger measures. Those not eligible or who
did not wish to participate in the experimental treatment were referred to other
programs. The intervention under investigation was administered as it would be in
clinical practice and participants paid a fee for the services they received. Fees were
set based on a standard group therapy fee schedule for the geographic area. The
clinic also provides a discounted fee schedule based on household income which is
 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279 265

frequently used to provide services to those who otherwise would be unable to afford
them. If the client wished, at the end of the project, a more standard anger
intervention would be provided at no additional cost.
Following the initial screening, a wait-list of eligible participants was created until
six people were available to form a treatment group. This resulted in a wait-list
period for three of the patients that ranged from 6 to 15 weeks. The three other
patients were screened within seven days of the start of the program. Wait-listed
clients completed all outcome measures during the initial screening and again at
sessions one and two. All the treatment sessions were conducted in a group format
by the first author. Once formed, the group met once a week for 90-min. In the case
of a missed group meeting, the patient simply returned the following week. Another
real world consideration was clinic resources. Therefore, the purpose of the group
meetings was to structure the intervention and to provide feedback to the patients.
The actual exposure practice trials were conducted by the patients at home. The first
author administered all assessment instruments.

1.3.1. Treatment outlineFsession 1

Patients introduced themselves and gave a brief overview of the history of their
anger problems. The importance of data collection was explained and the norm-
based measures of anger, anxiety, and depression were completed.

1.3.2. Session 2
The idiographic measures of anger were completed. Patients were guided through
a discussion of their most distressing ongoing anger situation. Client responses
produced specific details that could later be used in scene development. The therapist
introduced the concept of repeated exposure practice as a method of reducing
emotional arousal in response to anger triggers. Visualization was also introduced.
Clients were given several pleasant scenes to visualize in order to gain familiarity and
practice with these procedures.

1.3.3. Session 3
The therapist repeated the rationale for the systematic use of imaginal exposure.
Guidelines for scene development were provided and participants began to write
their scenes. In turn, each patient read his/her initial composition aloud while the
other group members practiced visualization. This was done to provide additional
visualization practice and also allow group members to give each other feedback
with regard to the detail and content. Homework was to rewrite the scenes, for the
next group meeting, incorporating the feedback received from other group members
and the therapist.

1.3.4. Session 4
The rationale for the exposure model was again repeated. The homework was
reviewed by having each patient in turn read his or her improved scene while the
other group members practiced imagining themselves in the situations presented.
Final suggestions were made for each of the scenes and remaining questions about
266 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279

imagery were addressed. The Daily Anger Exposure Record was then introduced
and explained. Once at home, patients were to record their scenes on an audiotape,
listen to the tape for one trial of at least 30-min per day, and complete the exposure
record. Since the use of exposure is new to the anger literature, and we were
uncertain how clients would react, it was decided to keep the exposure trial interval
to one half-hour.

1.3.5. Session 5
This session began with a review of the audiotape and Anger Exposure Records
for each patient. Each patient played his/her audiotape in order to receive feedback
from other group members and the therapist. Adjustments were made to each scene
as necessary. At this point in the program clients made modifications on their scenes
and engaged in exposure assignments based on their individual experiences and
progress. For example, one patient revised his tape to include more details, another
continued to review the original scene, and the four other patients created a variation
on their original scenes called ‘‘going to extremes’’. This required clients to create a
version that exaggerated negative events. All patients continued to review their
audiotaped anger scenes for one trial, 30 min each day, and to complete the Anger
Exposure Record.

1.3.6. Sessions 6–9

Sessions began with a review of the previous weeks records followed by a
discussion of experiences with the exposure practice. Again, depending on progress,
patients would revise scenes, make extreme scenes, or move on to second anger
situation. Continued exposure practice was assigned at the end of each meeting. Over
the course of treatment, each participant applied the exposure strategy to two anger
situations.

1.3.7. Session 10 (assessment)

The norm-based anger measures, idiographic anger measures, norm-based
measures of anxiety and depression, and the WAI were completed. Subjects
discussed their progress and future plans.
Fifteen months posttreatment, participants were sent the Trait Anger Scale, Anger
Expression Scales (anger-in, anger-out, anger-control) and a stamped return
envelope. In addition, a brief description of each participants’ initial anger situation
was provided, along with a rating scale, on which patients rated their current level of
anger intensity.

2. Results

2.1. Course of anger left untreated

As noted earlier, following the initial screening, three individuals experienced a

wait list period. Visual inspection of the baseline and pretreatment means on trait
 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279 267

anger (Mbaseline ¼ 21:67, SD¼ 1:15; Mpretreatment ¼ 26:33, SD ¼ 4:93), anger-in

(Mbaseline ¼ 20:67, SD ¼ 6:11; Mpretreatment ¼ 17:33, SD ¼ 4:51), anger-out
(Mbaseline ¼ 15:67, SD ¼ 3:51; Mpretreatment ¼ 17:67, SD ¼ 2:52), anger-control
(Mbaseline ¼ 19:00, SD ¼ 6:08; Mpretreatment ¼ 19:33, SD ¼ 6:81), anger situation
intensity (Mbaseline ¼ 80:00, SD ¼ 21:21; Mpretreatment ¼ 79:50, SD ¼ 27:58), and
anger situation frequency (Mbaseline ¼ 12:50, SD ¼ 16:26; Mpretreatment ¼ 30:00,
SD ¼ 18:21) indicate that left untreated, patients did not spontaneously remit and
that the course of the clinical anger problem tended to remain stable or worsen over
time. Exceptions are noted on measures of anger situation interference
(Mbaseline ¼ 80:00, SD ¼ 14:14; Mpretreatment ¼ 59:00, SD ¼ 55:86) and anger sensa-
tion intensity (Mbaseline ¼ 70:00, SD ¼ 42:43; Mpretreatment ¼ 57:59; SD ¼ 31:82),
which indicated some improvement. Mean baseline scores for anger situation
duration could not be calculated due to missing data.

2.2. Treatment effects and outcomes

Several approaches to judging the effectiveness of the intervention were employed.

The outcomes of these procedures along with the pretest and posttest means and
standard deviations are summarized in Table 1.
Statistically significant improvement was assessed for the sample as a whole with
the Wilcoxon Signed-ranks test. Due to the preliminary nature of this study, alpha
was set at po0:05 for all planned comparisons. At the end of treatment patients
reported significantly less trait anger, less anger-in, and fewer outward expressions of
anger (anger-out). While improvement is noted for controlled expression (anger-
control) these results did not reach statistical significance. In terms of the most
troublesome ongoing anger situation, significant reductions were reported for anger
intensity, anger duration, physiological arousal, and life interference. Improvements
on anger frequency were not significant. Reductions on anxiety and depression are
also noted, however, they did not reach significance. Overall, 11 hypotheses were
tested. The p-values obtained were rather small, many less than 0.05, and most less
than 0.10. This pattern suggests that patients experienced improvement. If patients
had not benefited from the program, more p-values in the range of 0.20–1.00 would
have been expected.
The magnitude of improvement over time was assessed for the entire sample by
calculating pre-to-posttest effect sizes (d) for each outcome measure. Effect sizes were
computed by dividing the difference between the pretreatment and posttreatment
means by the standard deviation of the pre to posttreatment difference scores.
According to Cohen (1977), for psychotherapy outcome research, effect sizes of 0.5
are considered moderate and those over 0.8 are considered large. Inspection of
Table 1 shows large treatment effects for all the anger measures, except anger
control, which yielded a moderate effect. A large treatment effect is also noted for
depression and a moderate effect was obtained on anxiety. While these results also
provide support for the effectiveness of the intervention, it should be noted that with
only six subjects effect sizes are sensitive to outliers.
268 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279

Table 1
Pre-to-posttest means, standard deviations, and outcomes for the 6 patientsa

Measures M SD Wilcoxon p-value Effect size No. of patients

Signed-ranks (one-tailed) (d) showing clinically
z-value significant
improvement

(a) Norm based anger measures

Trait anger
Pretest 27.00 4.56
Posttest 17.17 2.86 2.20 0.014 2.20 5
Anger-in
Pretest 18.67 3.39
Posttest 12.83 2.14 2.00 0.023 2.52 5
Anger-out
Pretest 22.00 6.10
Posttest 14.17 1.94 2.21 0.014 1.14 2
Anger control
Pretest 16.50 5.39
Posttest 20.83 4.96 1.36 0.089 0.57 3

(b) Idiographic anger measures

Anger situation intensity
Pretest 79.83 20.79
Posttest 15.83 15.63 2.20 0.014 2.06 5
Anger situation frequency
Pretest 14.00 14.03
Posttest 2.00 1.67 1.48 0.069 0.82 U
Anger situation duration
Pretest 27.50 19.94
Posttest 6.67 11.69 2.06 0.019 0.97 U
Anger situation interference
Pretest 61.50 44.74
Posttest 7.50 9.87 1.99 0.029 1.17 U
Anger sensation intensity
Pretest 71.67 31.09
Posttest 17.00 11.14 2.02 0.022 1.61 U

(c) Norm based measures of other emotional disturbance

Beck depression inventory
Pretest 15.33 12.86
Posttest 4.83 5.27 1.58 0.058 0.85 4
Beck anxiety inventory
Pretest 16.17 14.30
Posttest 6.17 5.19 0.94 0.173 0.60 3
a
Note. U=unable to compute due to the lack of normative data and reliability estimates. Non-clinical
norms for the BAI were not available in the technical manual and were obtained from Dent and
Salkovskis (1986).

Clinically significant improvement for each patient was evaluated according to the
criteria proposed by Jacobson and Truax (1991). Improvement is considered to be
clinically significant when (a) a patient’s score moves from the dysfunctional to the
 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279 269

functional range, and (b) when change is of sufficient magnitude to rule out random
fluctuations in measurement. In order to meet the first criteria, a patient’s posttest
score had to be closer to the M of a well functioning population (as identified in the
technical manual) than the dysfunctional population (pretreatment M for the
current sample). To meet the second criteria, a patient’s pre-to-posttest change
divided by the standard error of difference between the two scores had to exceed
1.96. The standard error of the difference was calculated by using the pretreatment
standard deviation and Cronbach’s coefficient alpha as the reliability estimate
(Tingey, Lambert, Burlingame, a Hansen, 1996). These analyses were not
performed on the idiographic measures, with the exception of anger intensity,
because normative data and reliability estimates were unavailable. Clinically
significant improvement was calculated for anger intensity using the Deffenbacher
et al. (1996) estimate of test-retest reliability and was achieved when a posttreatment
score moved at least two SDs, in the direction of improvement, from the
pretreatment M for the sample as a whole. It should be noted that there is some
debate regarding the conceptual usage and mathematical calculation of clinically
significant change. These issues are discussed in detail elsewhere (Follette a
Callaghan, 1996; Jacobson, Roberts, Berns, a McGlinchey, 1999; Kazdin, 1999;
Speer, 1992; Williams a Zimmerman, 1996).
Inspection of Table 1 reveals that on the norm-based anger measures, five of the
six patients showed clinically significant change on the trait anger and anger-in. Half
of the sample exhibited clinically significant improvement on anger control and two
patients moved into the normal distribution on anger-out. Five patients reported
clinically significant reductions in anger in response to their most serious ongoing
real life anger situation. In terms of non-anger measures, approximately half the
sample exhibited clinically significant decreases in symptoms of depression (four)
and anxiety (three).

2.3. Process measures

2.3.1. Habituation patterns

Fig. 1 shows anger intensity ratings after each trial of exposure practice for all 6
patients. Each box (two per participant) represents a distinct anger situation to
which exposure procedures were applied. The number of practice trials performed
for each situation and the timing of changes in scene content (revised or extreme)
varied from client to client. For imaginal exposure to be an active therapeutic
element responsible for the pre-to-posttreatment improvements, a pattern of
consistent and observable decreases on anger scores would be expected following
repeated exposure trials. Visual inspection of the graphs confirms this general
pattern, referred to as habituation.
Specifically, in response to the first situation, patient #1 showed an initial increase
in anger ratings followed by a steady decline over the course of 21 exposure trials.
For the second situation, each change in scene content resulted in an initial increase
in anger scores followed by immediate anger reduction. Patient #2 exhibited a similar
pattern for both situations. Anger ratings tended to increase with each change in
270 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279

scene content and decline with repeated rehearsal. Dramatic and rapid decreases in
anger ratings are evident for patient #3’s responses to the first situation. On the
second situation, anger ratings started very high and decreased over the course of 15
exposure trials resulting in consistent low ratings for the last five trials. Patient #4’s
results for the first anger situation were less positive. Although some improvement is
noted, his anger remained relatively high at trial 15. Further decreases may have
been achieved with more trials. A favorable response, however, was achieved on the
second situation. Although patient #5 attended the majority of group meetings, he
did not consistently record his at-home practice. This lack of documentation and
very low initial anger ratings, where high ratings were expected, brings into question

Fig. 1. Subjective ratings of anger intensity after each exposure practice trial. Boxes represent distinct
anger situations and breaks in the line indicate a shift to a revised or extreme version of the same situation.
 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279 271

Fig. 1. (continued).

the degree to which this patient followed the exposure protocol. Patient #6’s anger
ratings for the first situation were not very high. Nonetheless, decreases are still
evident over the course of 21 exposure trials. On the second situation, steady declines
were achieved over 24 trials. However, his anger ratings remained relatively high.
Further decreases might have been attained with more practice.
In addition to rating anger intensity at the end of each 30-min exposure trial,
patients also rated their anger 1-min into each trial. The M score for all practice
trials for all six subjects (based on 198 exposure records) at the 1-min interval was
3.53 and the M at the end of the trials was 3.13. This represents a within trial anger
decrease of less than one half a point on a 0–8 scale. Such a small per trial decrease
indicates that habituation effects tended to occur over repeated trials and not within
trials. Increasing the length of the exposure practice sessions, or having patients rate
272 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279

their anger at its peak, rather than after 1-min, might have resulted in greater
reported within trial anger reductions.

2.3.2. Patient satisfaction

Overall, session attendance was good. Four patients attended all 10 meetings.
Patient #5 missed two of the sessions and patient #1 missed one session. At the end
of the exposure protocol, the group leader offered to set up a more traditional
cognitive-behavioral program for those who believed that they did not benefit
substantially. All 6 patients reported satisfaction with the current program and did
not wish to pursue another intervention.
On the WAI, ratings of the perceived alliance between each participant and the
group leader ranged from 5.33 to 6.83 (M ¼ 6:40; total score). Ratings for agreement
on goals ranged from 5.75 to 6.83 (M ¼ 6:39), agreements on tasks 4.68 to 7.00
(M ¼ 6:40), and formation of a bond 5.88 to 6.75 (M ¼ 6:47). Scores of 4.5 or less
are considered to indicate problems in the therapeutic alliance and predict premature
termination (Samstag et al., 1998). Scores from the current project are high (all
over 4.5) and indicate an acceptable alliance with the group leader and a positive
treatment experience for each patient.

2.4. Long-term outcomes

Three patients (50%) returned data at 15-month follow-up. The three non-
respondents had either moved out of the geographic area or failed to reply to
repeated mailings. A series of Mann-Whitney U tests were conducted on pretest and
posttest scores to explore possible differences between patients who provided follow-
up data and those who did not. No significant differences were found, indicating that
patients who provided follow-up information were representative of the sample as a
whole in terms of their initial anger scores and their responses to treatment. Visual
inspection of the means for the three patients appears to indicate sustained
improvement on all anger measures for which follow-up data were collected (see
Table 2). Statistically significant results were maintained for trait anger, anger-out,
and anger situation intensity. Results for anger-in and anger control were non-
significant. Effect sizes indicate that moderate to large improvements were
maintained and the majority of patients continued to show clinically meaningful
and reliable change on all measures except anger-out.

3. Discussion

This investigation was an initial attempt to apply an imaginal exposure strategy to

angry clients. It represents one of many possible ways of using exposure procedures
to reduce anger. The present investigation differs from most others in the anger
treatment literature in that the participants were actual clinic referrals and the
treatment was administered, as it would be in real clinical practice. Although most of
our patients had histories of aggressive behavior, the focus of the present program
 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279 273

Table 2
Pretest, posttest, and fifteen-month follow-up means, standard deviations, and outcomes for 3 patientsa

Measures M SD Friedman p-value Effect No. of patients

test X 2 ð2Þ (one-tailed) size ðdÞ showing clinically
significant
improvement

Trait anger
Pretest 29.00 5.19
Posttest 18.67 3.52
Follow-up 19.67 3.06 5.64 0.030 1.69 2
Anger-in
Pretest 18.00 4.58
Posttest 13.33 2.31
Follow-up 14.00 3.46 0.55 0.381 0.64 2
Anger-out
Pretest 20.67 7.37
Posttest 13.67 2.89
Follow-up 14.33 1.53 4.67 0.048 0.84 1
Anger control
Pretest 17.67 8.14
Posttest 22.33 4.93
Follow-up 23.33 1.53 0.68 0.359 0.67 2
Anger situation intensity
Pretest 86.33 22.81
Posttest 13.33 15.28
Follow-up 6.67 11.55 5.64 0.030 3.98 3
a
Note. Clinically significant improvement and d’s based on comparisons between pretest and follow-up
scores.

was to reduce the subjective (internal) experience of anger. Thus, measurements were
obtained on a variety of self-report dimensions including both normative and person
specific outcomes. The impact of the intervention was judged from several
perspectives. Statistically significant change and magnitude of change were assessed
at the group level and clinically significant improvement was evaluated for each
individual. Process data in response to exposure practice trials were also obtained for
each patient. In addition, long-term treatment effects were examined for several
patients.
In spite of the small sample size, statistically significant change was reached on
seven out of the nine anger measures. On normative measures, patients showed
decreases in trait anger, anger-in, and anger-out. In regards to a personal and
provocative ongoing situation, patients reported lowered anger intensity, shorter
duration of reported angry feelings, fewer physical sensations, and less interference
with everyday functioning. While there appeared to be some generalization to other
emotional problems (e.g. anxiety and depression), these results were not significant
at the group level.
The magnitude of change also appeared clinically meaningful. Treatment effect
sizes were in the large range on all anger variables except anger control. A large
274 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279

treatment effect was also found for depression and a moderate effect for anxiety. A
number of other interventions studied in the anger treatment outcome literature have
also yielded large effect sizes.2 For example, using trait anger as a comparison
variable, self-instructional training obtained a pre-to-post effect size of 1.31
(Deffenbacher, Story, Brandon, Hogg, a Hazaleus, 1988), Beck’s cognitive therapy
1.73 (Deffenbacher et al., 2000), social skills training 0.99 (Deffenbacher et al., 1994),
problem solving 1.31 (Deffenbacher et al. (1994)), group process 1.17 (Deffenbacher,
McNamara, Stark, a Sabadell, 1990), anger management training 2.39 (Deffenba-
cher a Stark, 1992), and a combined cognitive-relaxation intervention 1.82
(Deffenbacher a Stark, 1992). The current program, with an effect size of 2.17
for trait anger, compares reasonably well to other interventions. The only
intervention that produced a higher pre-to-posttreatment effect size was Anger
Management Training (AMT; Deffenbacher a Stark, 1992) which also utilizes
imaginal exposure procedures.
At the individual level, clinical significance was attained when a patient’s posttest
score was not distinguishable from a non-disturbed reference group and when the
change was of a sufficient magnitude to be reliable. The majority of participants
(five) showed a degree of improvement that would be classified as clinically
significant on three out of the five anger measures for which this criterion was
applied. This occurred for the variables of trait anger, anger-in, and anger in
response to the most troublesome real life situation. Half the sample (three patients)
showed movement into the functional range for controlled expressions of anger
(anger-control) and only two reached clinical significance on anger-out. The test may
not have been appropriate for the anger-out variable given high degree of variability
at pretest resulting in a large standard error of measurement. Four patients reported
clinically significant change on depressive symptoms and three for anxiety related
symptoms.
In considering the various methods of assessing improvement, imaginal exposure
appeared to be an effective intervention for the majority of the patients on many
important indices of anger. As expected, the intervention seemed to have less impact
on symptoms of anxiety and depression. Nonetheless, some degree of improvement
is noted in these areas, which is consistent with other reports found in the treatment
outcome literature. One potential explanation is that anger reduction contributed to
better overall functioning resulting in fewer triggers for anxiety and depression.
The process data suggests that one patient (subject #5) did not adhere to the
exposure protocol. An inspection of the outcome data for this patient revealed that
he failed to achieve clinical significance on most of the anger measures. Given that
there were only 6 participants, this tainted the group results somewhat. Nonetheless,
this case raises an important treatment issue. The current program required active

2
For comparison purposes, pre-to-posttest effect sizes (d) were calculated by the authors, from data
provided in published reports. Effect sizes were computed by dividing the difference between the
pretreatment and posttreatment mean by the pooled standard deviation of both scores. These calculations
were performed with the D-STAT program (Johnson, 1989), which also provides an adjustment for sample
size.
 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279 275

daily participation from each client. Exposure procedures may be best suited for
those individuals who are sufficiently motivated and willing to comply with between
session assignments. As noted by other researchers, building motivation for change
may be a critical issue when working with angry adults (DiGiuseppe, Tafrate, a
Eckhardt, 1994).
For the remaining five participants, process data indicated consistent and
observable decreases in anger on 9 of the 10 situations for which scenes were
created and rehearsed. Only patient #4 failed to show a reduction in anger for the
first situation. Although baseline data for each participant would have been useful
for interpreting change and ruling out extraneous variables, given the real world
concerns of the patients, treatment was not delayed for this purpose. As noted
earlier, baseline data on anger outcome measures, including the most serious
ongoing anger situation, from which the first anger exposure scenes were developed,
was obtained from three clients awaiting treatment. This data provided some basic
information regarding the course of clinical anger in the present sample and revealed
that anger situation intensity ratings remained stable and unchanged after several
weeks of no treatment. Once the intervention was applied, decreases in anger ratings
for the first situation were noted. Exposing patients to a second anger situation then
replicated these results. The overall pattern of change is consistent with a habituation
effect and is supportive of the role of exposure as an active therapeutic element.
Based on a visual inspection of the graphs, it appears that the process of habituation
in angry clients is similar to what has been observed in clients with anxiety disorders.
Several other patterns emerged from the process data. All patients who adhered to
the protocol showed an initial increase in anger when moving from the first anger
situation to the second. Generalization between situations may not occur naturally
and therefore it may be important to repeatedly expose clients to all of their major
anger triggers. In addition, some clients exhibited initial increases when confronting
a revised or extreme version of the same situation. Such increases generally appeared
short-lived and decreases were noted within one or two trials. In the present
program, decisions about when to make a change in scene content or situation were
client driven. Patient boredom was often the main reason for making a shift. A
stronger habituation effect might have been achieved by continuing with the same
scene for more trials. Perhaps a better way to guide decisions about shifts to new
scenes is to utilize an empirical criterion, such as three trials with anger ratings of less
than two.
Average anger ratings within trials indicated that very little change occurred
during the one half-hour exposure practice sessions. It appears that habituation
occurred over repeated trials but not within trials. The trial length may have been
too short for anger arousal to subside, and greater within trial decreases may have
been achieved with a massed practice approach (Levis, 1980). Small within trial
habituation effects might also be due to way in which measurements were recorded.
Rather than have patients rate their anger 1-min into a trial, larger decreases might
have been observed had patients rated their anger at its peak. Further empirical
work should focus on evaluating the habituation process in terms of optimal trial
length and number of rehearsals.
276 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279

Participant satisfaction with the present program was high as suggested by

good attendance and WAI scores. Patients appeared to understand the rationale
for engaging in repeated exposure practice and did not report any adverse
affects. Although patients had a variety of anger triggers, imaginal exposure
procedures were easily adapted to a group format. The role of the group leader
in the present program was to obtain assessment information and to provide
structure and feedback. Patients conducted most of the exposure practices
on their own outside of the group sessions. It is possible that this interven-
tion could be structured in a manual and presented in a self-help format. Addi-
tional research with other types of clients with anger problems would be
useful in identifying any contra-indications for the use of exposure
procedures.
Finally, statistically significant and clinically meaningful change was evident at 15-
months following the intervention. In comparing posttest and follow-up scores
(Tables 1 and 2), statistically significant improvement was maintained on trait anger,
anger-out, and anger situation intensity. Although effect sizes decreased on several
measures, gains were still in the moderate to large range at the end of the follow-up
period. As was the case at posttest, clinically significant and reliable change was also
observed at follow-up for the majority of patients on all measures except anger-out.
Persistence of treatment effects was observed on both normative and person specific
measures. Patients reported continued improvement for the situations that were
specifically addressed in the exposure program, as well as for general anger
experiences.
While administering the treatment in a community mental health center to actual
anger referrals added to the ecological validity of this investigation, it also
contributed to a number of limitations. Our methodology did not allow for a
comparison between patients treated with exposure to a similarly referred group of
anger patients that received no treatment or some other intervention, for the same
period of time. Therefore, the present findings should be interpreted with caution. A
supportive relationship among the patients, praise and positive feedback, increased
awareness of anger reactions, and wanting to please the therapist-assessor may have
all contributed to the observed improvements. Replication of the present results is
certainly needed. Better experimental control is required to rule out a number of
important threats to internal validity inherent in this pilot project and allow for
stronger causal inferences.
In spite of the obvious limitations, the present study indicates that imaginal
exposure can be applied to adults with clinical anger problems. Preliminary results
provide encouraging suggestive evidence that imaginal exposure may be an effective
intervention. The development of a comprehensive exposure model of anger
treatment requires a far greater understanding of the process of habituation in angry
clients than now exists. Future researchers may wish to investigate the relationship
between exposure trial length and habituation, optimal number of trials required for
anger reduction to a specific stimulus, and generalization across anger situations.
Exposure-based procedures certainly warrant further exploration as an anger
reduction strategy.
 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279 277

References

Barlow, D. H., a Craske, M. G. (1994). Mastery of your anxiety and panic (client workbook), (2nd ed.).
San Antonio, TX: The Psychological Corporation.
Beck, A. T., Steer, R. A., a Brown, G. K. (1996). Beck Depression Inventory, (2nd ed.). San Antonio, TX:
The Psychological Corporation.
Beck, A. T., a Steer, R. A. (1993). Beck Anxiety Inventory. San Antonio, TX: The Psychological
Corporation.
Brondolo, E., DiGiuseppe, R., a Tafrate, R. (1997). Exposure-based treatment for anger problems:
Focus on the feeling. Cognitive and Behavioral Practice, 4, 75–98.
Cohen, J. (1977). Statistical power analysis for the behavioral sciences. New York: Academic Press.
Cornell, D. G., Peterson, C. S., a Richards, H. (1999). Anger as a predictor of aggression in incarcerated
adolescents. Journal of Consulting and Clinical Psychology, 67, 108–115.
Craske, M. G., Barlow, D. H., a O’Leary, T. (1992). Mastery of your anxiety and worry (client
workbook). San Antonio, TX: The Psychological Corporation.
Deffenbacher, J. L. (1992). Trait anger: Theory, findings, and implications. In C. D. Spielberger, a J. N.
Butcher, Advances in personality assessment, vol. 9 (pp. 177–201). Hillsdale, NJ: Erlbaum.
Deffenbacher, J. L., Dahlen, E. R., Lynch, R. S., Morris, C. D., a Gowensmith, W. N. (2000).
Application of Becks cognitive therapy to general anger reduction. Cognitive Therapy and Research, 24,
689–697.
Deffenbacher, J. L., Demm, P. M., a Brandon, A. D. (1986). High general anger: Correlates and
treatment. Behaviour Research and Therapy, 24, 481–489.
Deffenbacher, J. L., McNamara, K., Stark, R. S., a Sabadell, P. M. (1990). A comparison of cognitive-
behavioral and process-oriented group counseling for general anger reduction. Journal of Counseling
and Development, 69, 167–172.
Deffenbacher, J. L., Oetting, E. R., Lynch, R. S., a Morris, C. D. (1996). The expression of anger and its
consequences. Behaviour Research and Therapy, 34, 575–590.
Deffenbacher, J. L., a Stark, R. S. (1992). Relaxation and cognitive-relaxation treatments of general
anger. Journal of Counseling Psychology, 39, 158–167.
Deffenbacher, J. L., Story, D. A., Brandon, A. D., Hogg, J. A., a Hazaleus, S. L. (1988). Cognitive and
cognitive-relaxation treatments of anger. Cognitive Therapy and Research, 12, 167–184.
Deffenbacher, J. L., Thwaites, G. A., Wallace, T. L., a Oetting, E. R. (1994). Social skills and
cognitive-relaxation approaches to general anger reduction. Journal of Counseling Psychology, 41,
386–396.
DeMoja, C. A., a Spielberger, C. D. (1997). Anger and drug addiction. Psychological Reports, 81, 152–
154.
Dent, H. R., a Salkovskis, P. M. (1986). Clinical measures of depression, anxiety, and obsessionality in
non-clinical populations. Behaviour Research and Therapy, 24, 689–691.
DiGiuseppe, R., Tafrate, R., a Eckhardt, C. (1994). Critical issues in the treatment of anger. Cognitive
and Behavioral Practice, 1, 111–132.
Dua, J. K., a Swinden, M. L. (1992). Effectiveness of negative thought reduction, meditation,
and placebo training treatment in reducing anger. Scandinavian Journal of Psychology, 33,
135–146.
Edmondson, C. B., a Conger, J. C. (1996). A review of treatment efficacy for individuals with anger
problems: Conceptual, assessment, and methodological issues. Clinical Psychology Review, 16, 251–
275.
Ellis, A. (1962). Reason and emotion in psychotherapy. New York: Citadel Press.
Evans, D. R., Hearn, M. T., a Saklofske, D. (1973). Anger, arousal, and systematic desensitization.
Psychological Reports, 32, 625–626.
Foa, E. B., a Rothbaum, B. A. (1998). Treating the trauma of rape: Cognitive behavioral therapy for
PTSD. New York: Guilford.
Foa, E. B., a Wilson, R. (1991). Stop obsessing: How to overcome obsessions and compulsions. New York:
Bantam Books.
278 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279

Follette, W. C., a Callaghan, G. M. (1996). The importance of the principle of clinical significance -
Defining significant for whom and for what purpose: A response to Tingey, Lambert, Burlingame, and
Hansen. Psychotherapy Research, 6, 133–143.
Hazaleus, S. L., a Deffenbacher, J. L. (1986). Relaxation and cognitive treatments of anger. Journal of
Consulting and Clinical Psychology, 54, 222–226.
Helmers, K. F., Posluszny, D. M., a Krantz, D. S. (1994). Associations of hostility and coronary heart
disease: A review of studies. In A. Siegmen, a T. Smith (Eds.), Anger, hostility, and the heart. Hillsdale
NJ: Lawrence Erlbaum Associates.
Horvath, A. O., a Greenberg, L. S. (1989). Development and validation of the Working Alliance
Inventory. Journal of Counseling Psychology, 36, 223–11233.
Jacobson, N. S., Gottman, J. M., Waltz, J., Rushe, R., Babcock, J., a Holtzworth-Monroe, A. (1994).
Affect, verbal content, and psychophysiology in the arguments of couples with a violent husband.
Journal of Consulting and Clinical Psychology, 62, 982–988.
Jacobson, N. S., Roberts, L. J., Berns, S. B., a McGlinchey, J. B. (1999). Methods for defining and
determining the clinical significance of treatment effects: Description, application, and alternatives.
Journal of Consulting and Clinical Psychology, 67, 300–307.
Jacobson, N. S., a Truax, P. (1991). Clinical significance: A statistical approach to defining meaningful
change in psychotherapy research. Journal of Consulting and Clinical Psychology, 59, 12–19.
Johnson, B. T. (1989). DSTAT: Software for the meta-analytic review of research literatures. Hillsdale, NJ:
Erlbaum.
Kassinove, H., a Sukhodolsky, D. G. (1995). Anger disorders: Basic science and practical issues. In H.
Kassinove (Ed.), Anger disorders Definition diagnosis and treatment (pp. 1–11). Washington DC: Taylor
and Francis.
Kazdin, A. E. (1999). The meaning and measurement of clinical significance. Journal of Consulting and
Clinical Psychology, 67, 332–339.
Kroner, D. G., a Reddon, J. R. (1995). Anger and psychopathology in prison inmates. Personality and
Individual Differences, 18, 783–788.
Lachmund, E., a DiGiuseppe, R. (1997). How clinicians assess anger: Do we need an anger diagnosis? In
R. DiGiuseppe (Chair), Advances in the diagnosis, assessment, and treatment of angry clients.
Symposium conducted at the 105th annual convention of the American Psychological Association,
August, Chicago, IL: American Psychological Association.
Levis, D. J. (1980). Implementing the technique of implosive therapy. In A. Goldstein, a E. B. Foa (Eds.),
Handbook of behavioral interventions: A clinical guide. New York: John Wiley and Sons.
Lopez, F. G., a Thurman, C. W. (1986). A cognitive behavioral investigation of anger among college
students. Cognitive Therapy and Research, 10, 245–256.
Markway, B. G., Carmin, C. N., Pollard, C. A., a Flynn, T. (1992). Dying of embarrassment: Help for
social anxiety and phobia. Oakland, CA: New Harbinger Publications.
O’Donnell, C. R., a Worell, L. (1973). Motor and cognitive relaxation in the desensitization of anger.
Behaviour Research and Therapy, 11, 473–481.
Rimm, D. C., DeGroot, J. C., Boord, P., Heiman, J., a Dillow, P. V. (1971). Systematic desensitization of
an anger response. Behaviour Research and Therapy, 9, 273–280.
Samstag, L. W., Batchelder, S. T., Muran, J. C., Safran, J. D., a Winston, A. (1998). Early identification
of treatment failures in short-term psychotherapy: An assessment of therapeutic alliance and
interpersonal behavior. Journal of Psychotherapy Practice and Research, 7, 126–143.
Speer, D. C. (1992). Clinically significant change: Jacobson and Truax (1991) revisited. Journal of
Consulting and Clinical Psychology, 60, 402–408.
Spielberger, C. D. (1988). State-Trait Anger Expression Inventory Odessa, FL: Psychological Assessment
Resources Inc.
Suinn, R. M. (1990). Anxiety management training: A behavior therapy. New York: Plenum Press.
Tafrate, R. (1995). Evaluation of treatment strategies for adult anger disorders. In H. Kassinove (Ed.),
Anger disorders: Definition, diagnosis, and treatment. Washington DC: Taylor and Francis.
Tafrate, R., DiGiuseppe, R., a Parsi, F. (1997). A review of treatment efficacy for adult anger disorders:
Published studies, dissertations, and related investigations. In R. DiGiuseppe (Chair), Advances in the
 G.R. Grodnitzky, R.C. Tafrate / J. Behav. Ther. & Exp. Psychiat. 31 (2000) 259–279 279

diagnosis, assessment, and treatment of angry clients. Symposium conducted at the 105th annual
convention of the American Psychological Association, August, Chicago, IL: American Psychological
Association.
Tafrate, R., a Kassinove, H. (1998). Anger control in men: Barb exposure with rational, irrational, and
irrelevant self-statements. The Journal of Cognitive Psychotherapy, 12, 187–211.
Tingey, R. C., Lambert, M. J., Burlingame, G. M., a Hansen, N. B. (1996). Assessing clinical
significance: Proposed extensions to method. Psychotherapy Research, 6, 109–123.
Wallfish, S., Massey, R., a Krone, A. (1990). Anxiety and anger among abusers of different substances.
Drug and Alcohol Dependence, 25, 253–256.
Whiteman, M., Fanshel, D., a Grundy, J. F. (1987). Cognitive-behavioral interventions aimed at anger of
parents at risk for child abuse. Social Work, 32, 469–474.
Williams, R. H., a Zimmerman, D. W. (1996). Are simple gain scores obsolete? Applied Psychological
Measurement, 20, 59–69.